Multimodal Analytical Platform Coupled With Biological Assays Unravels Tissue‐Specific Chemical Profiles, Spatial Metabolite Distributions and Bioactivities of Root Xylem and Root Bark of Tripterygium wilfordii

ABSTRACT

Characterising chemical and biological differences among different tissues of medicinal plants is important for understanding their tissue‐specific characteristics and medicinal efficacy. The root xylem (RX) and root bark (RB) of Tripterygium wilfordii have historically been used differently; however, the basis underlying their differential utilisation remains insufficiently characterised. In this study, complementary analytical measurements were integrated to compare the chemical and biological differences between the two root tissues across multiple dimensions. Ion fingerprinting revealed substantial overlap in the detected chemical features, with 81.09% of the ion features in the RX also detectable in the RB. Despite this high proportion of shared constituents, multivariate statistical analysis clearly separated the RX and RB, indicating significant differences in their overall ion abundance patterns. Mass spectrometry imaging further resolved tissue‐specific distribution patterns, showing that alkaloids were mainly localised in the phloem, triterpenoids were enriched in the periderm, whereas diterpenoids exhibited broader distributions across the root tissues. Targeted quantitative analysis provided direct evidence for the differential abundance of metabolites, showing that multiple bioactive constituents, including triptolide, triptonide and celastrol, were present at considerably higher levels in the RB than in the RX. Biological evaluation further demonstrated that, compared with the RX, the RB exhibited stronger inhibition of lipopolysaccharide‐induced nitric oxide production, albeit with higher cytotoxicity. Taken together, the complementary measurements confirmed that, although the RX and RB share a large number of chemical features, they differ markedly in overall metabolite abundance and spatial distribution, ultimately exhibiting distinct biological responses. This study established a multidimensional analytical framework for resolving tissue‐specific chemical variation in complex medicinal plant materials and demonstrates its utility for integrating chemical profiling, spatial mapping, targeted quantification and biological evaluation.

Abbreviations

AC

active control

CCK‐8

cell counting kit‐8

DHB

2,5‐dihydroxybenzoic acid

DMSO

dimethyl sulfoxide

FBS

foetal bovine serum

HCA

hierarchical clustering analysis

HCD

high‐energy collision dissociation

ITO

indium tin oxide

LPS

lipopolysaccharide

MRM

multiple reaction monitoring

MSI

mass spectrometry imaging

NO

nitric oxide

PCA

principal component analysis

RB

root bark

RX

root xylem

TICs

total ion chromatograms

UHPLC‐QQQ‐MS/MS

ultra‐high‐performance liquid chromatography‐tandem triple quadrupole mass spectrometry

1. Introduction

Medicinal plants are often used as different tissues or anatomical parts, which may have distinct traditional applications despite being derived from the same plant species. Understanding the chemical and biological basis underlying such tissue‐specific utilisation is important for elucidating the material basis of traditional use and for supporting the rational utilisation of medicinal plant resources. However, differences between medicinal tissues cannot be fully understood from individual marker compounds alone, as differences in overall chemical composition, metabolite abundance, tissue localisation and biological properties may collectively contribute to their distinct applications. A comprehensive characterisation therefore requires multiple lines of chemical and biological evidence to understand these differences from complementary perspectives.

Tripterygium wilfordii Hook. f. is a well‐known medicinal plant that has been used in traditional Chinese medicine for centuries, particularly for the treatment of rheumatoid arthritis and other autoimmune diseases [1, 2]. Its roots are the principal medicinal part and are commonly divided into root xylem (RX) and root bark (RB, comprising the phloem and periderm). These two tissues have historically been used differently, with RX more commonly used for internal administration and RB primarily used externally [3, 4, 5]. A similar distinction exists in modern pharmaceutical production: standardised preparations such as Tripterygium glycosides tablets are predominantly prepared from RX, whereas RB is generally removed during processing [6, 7]. Previous studies have compared selected chemical constituents between RX and RB, particularly representative diterpenoids and triterpenoids such as triptolide and celastrol and have reported differences in their abundances [8, 9, 10]. These findings suggest that chemical differences may contribute to the distinct utilisation of the two tissues. However, these targeted comparisons provide only a limited view of their overall chemical characteristics and do not clarify their tissue‐specific chemical distribution or associated biological properties.

Mass spectrometry (MS)‐based measurements provide complementary information for characterising these aspects of chemical variation. Ion fingerprinting enables comprehensive comparison of chemical features across samples, while multivariate analysis can further identify the major variables contributing to differences between tissues [11, 12]. MS imaging (MSI) provides spatial information on metabolite localisation, and targeted quantitative analysis allows differences in the abundance of representative constituents to be directly measured [13]. Together, these measurements provide complementary information on the chemical features, metabolite abundance and spatial distribution of plant tissues.

In this study, we applied an integrated analytical and biological approach to compare RX and RB of T. wilfordii. Ion fingerprinting and multivariate analysis were first used to characterise their overall chemical features and abundance patterns. Selected constituents were subsequently investigated by MSI to determine their tissue‐specific localisation and by targeted quantitative analysis to characterise differences in metabolite abundance. These chemical differences were further evaluated in relation to the cytotoxic and anti‐inflammatory properties of the two tissues. By integrating global chemical profiling, spatial molecular mapping, targeted quantification and biological evaluation, this study aims to clarify the chemical and biological basis underlying the differential utilisation of RX and RB and to provide a more comprehensive understanding of the tissue‐specific characteristics of T. wilfordii roots.

2. Materials and Methods

2.1. Materials and Reagents

Triptolide, celastrol, triptonide, triptophenolide, wilforlide A, indomethacin, blinin, evodiamine and glycyrrhetic acid were purchased from the National Institutes for Drug Control (Beijing, China). Wilforgine, wilforine, peritassine A, pristimerin, demethylzeylasteral, triptoquinone B, triptoquinone A, wilfortrine and wilfordine were purchased from Shanghai Yuanye Biotechnology Co., Ltd (Shanghai, China). The chemical purity of all reference standards was not less than 98% as determined by high‐performance liquid chromatography (HPLC).

Methanol and acetonitrile were purchased from Honeywell (Charlotte, NC, USA). Formic acid for LC‐MS grade was purchased from Waters (Milford, MA, USA). Dimethyl sulfoxide (DMSO), lipopolysaccharide (LPS) and 2,5‐dihydroxybenzoic acid (DHB) were purchased from Sigma‐Aldrich (St. Louis, MO, USA). Indium tin oxide (ITO) conductive glass was purchased from Matsunami Glass Ind., Ltd. (Osaka, Japan). The RAW 264.7 (mouse macrophage, male, CVCL_0493), GC‐2spd(ts) (mouse spermatocyte, male, CVCL_6633) and Hep‐G2 (human hepatocellular carcinoma, male, CVCL_0027) cell lines were obtained from Wuhan Procell (Wuhan, China). DMEM medium and foetal bovine serum (FBS) were purchased from Yeasen Biotechnology Co., Ltd. (Shanghai, China). RPMI‐1640 medium was purchased from Analysis Quiz (Shanghai, China). The cell counting kit‐8 (CCK‐8) was purchased from Meilunbio (Dalian, China); the nitric oxide (NO) assay kit was obtained from Applygen Technologies Inc. (Beijing, China).

2.2. Sample Collection, Identification and Categorisation

The roots and aboveground parts of T. wilfordii were collected from Zhejiang (ZJ), Fujian (FJ), Hubei (HB) and Hunan (HN) provinces, respectively, as shown in Table S1. Morphological identification was performed by Shuai Kang (National Institutes for Drug Control). Molecular identification was conducted by Kai Ling Xu (Institute of Materia Medica, Chinese Academy of Medical Sciences and Peking Union Medical College) and corroborated the morphological classification [14]. The dried roots were manually dissected along the natural boundary between the bark (phloem and periderm) and xylem tissues. The separated bark fraction was designated as RB, while the remaining woody tissue was designated as RX. Finally, 20 batches of RX and RB were obtained, respectively. Then the samples were powdered and filtered through a 65‐mesh sieve.

2.3. Sample Preparation and Analysis Conditions for Ion Fingerprinting

0.2 g of sample was ultrasonically extracted with 20 mL of methanol at room temperature for 45 min. After extraction, the extract was allowed to cool to room temperature, and methanol was added to compensate for solvent loss and restore the extract to its initial weight. The extract was then thoroughly mixed and filtered through a 0.22 µm microporous membrane before analysis. A quality control sample was prepared by mixing equal volumes of all samples for system stability monitoring. Data were acquired using a Vanquish UHPLC system coupled with an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scientific, MA, USA). Chromatographic separation was achieved on an Agilent ZORBAX Extend C18 column (2.1 × 100 mm, 1.8 µm). The mobile phase consisted of solvent A (0.1% formic acid in water) and solvent B (acetonitrile), with the following gradient elution program: 0–40 min, 95%–5% A; 40–45 min, 5% A; 45–50 min, 5%–95% A; The flow rate was set at 0.3 mL/min, the injection volume was 3 µL and the column temperature was maintained at 35°C.

Mass spectrometric data were acquired in positive ion mode using a heated electrospray ionisation source. Parameters were as follows: sheath gas, 45 psi; auxiliary gas, 15 psi; ion transfer tube temperature, 350°C; vaporiser temperature, 320°C; spray voltage, 3500 V. The Orbitrap analyser scanned over a mass range of m/z 100–1500 at a resolution of 120,000 for full‐scan MS [1] and a resolution of 30,000 for ddMS² in a profile format with high‐energy collision dissociation (HCD) mode. HCD energy was 40% and 60% in stepped mode.

Raw data were processed using Compound Discoverer 3.3 for peak detection, alignment and feature extraction. The mass tolerance and retention time tolerance were set to 5 ppm and 0.1 min, respectively, while other processing parameters were maintained at the default settings of the software. The resulting feature‐ion intensity matrix was imported into MetaboAnalyst 6.0 for statistical analysis. Principal component analysis (PCA) was performed using the original feature‐ion intensity matrix comprising all detected feature ions, without mean centring, scaling, or z‐score normalisation. Hierarchical clustering analysis (HCA) was performed using the same feature‐ion intensity matrix, with Euclidean distance as the distance measure and Ward's method as the linkage method. The feature‐ion data obtained from Compound Discoverer 3.3 were further processed using Python 3.13 to generate ion fingerprinting and calculate the ion coverage between RB and RX. The ion coverage was calculated using the following equation:

$$Ioncoverage=\frac{RB\cap RX}{RX}\times 100\%$$

2.4. Sample Preparation and Analysis Conditions for MSI

Tissue segments measuring approximately 1–2 cm in length were obtained from the central region of fresh roots of T. wilfordii, embedded in 10% (w/v) gelatin and cryopreserved at −80°C overnight. Cryosectioning was performed using a Leica CM1950 cryostat, with the chamber temperature set to −20°C and the blade maintained at −14°C. Transverse sections were cut at a thickness of 20 µm and carefully mounted onto ITO‐coated conductive glass slides. A DHB matrix solution was then uniformly applied using an automated sprayer. The spraying protocol consisted of 25 passes at a flow rate of 120 µL/min, with a stage speed of 70 mm/s and a nozzle‐to‐sample distance of 10 cm.

Chemical reference standards were prepared by dissolving the compounds in methanol to a final concentration of 0.5 mg/mL. Aliquots of these standard solutions were spotted onto ITO‐coated glass slides and air‐dried, followed by matrix application using the same automated spraying method. These preparations were used to determine the optimal adduct ions for subsequent analysis.

Matrix‐assisted laser desorption/ionisation‐MSI platform (iMScope QT) equipped with an LCMS‐9030 Q‐TOF mass analyser (Shimadzu, Tokyo, Japan) was used to obtain spatial distribution profiles of metabolites in root tissue. Acquisition parameters were set as follows: laser diameter, 25 µm; laser intensity, 75 au; laser frequency, 1000 Hz; and step size, 50 µm. Data were acquired in positive ion mode with an ionisation voltage of 2.1 kV and a mass range of m/z 100–1000. Data were processed and visualised using ImageReveal MS software.

2.5. Sample Preparation and Optimisation of Analysis Conditions for Targeted Quantitative Analysis

Sample preparation was performed using the same procedure as that described for ion fingerprinting. The sample extracts were diluted 2‐, 50‐ and 10‐fold for the determination of diterpenoids, triterpenoids and alkaloids, respectively. Distinct internal standards, blinin (IS1), glycyrrhetic acid (IS2) and evodiamine (IS3), were used for diterpenoids, triterpenoids and alkaloids, respectively. Data was carried out by Waters Xevo TQ Absolute systems equipped with an ACQUITY UPLC H‐Class PLUS system (Waters, Milford, MA, USA). Agilent ZORBAX Extend C18 (2.1 × 100 mm, 1.8 µm) was used for separation. The mobile phase was composed of (A) 0.1% formic acid in water and (B) 0.1% formic acetonitrile. Three types of components were subjected to distinct gradient elution procedures due to the significant polarity difference. For alkaloids: 0–25 min, 75%–30% A; 25–27 min, 30%–5%A; 27–32 min, 5%A. For diterpenoids: 0–8 min, 95%–85% A; 8–12 min, 85%–75%A; 12–25 min, 75%–35% A;25–27 min, 35%–5% A; 27–33 min, 5% A. For triterpenoids: 0–10 min, 50%–35% A; 10–20 min, 35%–5%A; 20–25 min, 5% A. The flow rate was 0.3 mL/min, the injection volume was 2 µL, and the column temperature was maintained at 35°C. Mass data were acquired in multiple reaction monitoring (MRM) mode. Compound information and acquisition parameters were shown in Table S2. This method had already been fully validated in our previous studies, including evaluation of calibration curves, limit of detection/limit of quantitation, precision, stability, repeatability and recovery [15]. Data visualisation was performed using GraphPad Prism.

2.6. Preparation and Procedures for In Vitro Anti‐Inflammatory Activity and Toxicity Assays

A 0.5 g sample was extracted with 50 mL of methanol by ultrasonication for 45 min at room temperature. Then the extract was evaporated to dryness and redissolved in 1 mL of DMSO. The extract was evaporated to dryness and redissolved in 1 mL of DMSO. For the anti‐inflammatory assay, RAW 264.7 cells in the logarithmic growth phase were seeded into 96‐well plates at a density of 3 × 10⁴ cells/well in DMEM supplemented with 10% FBS and incubated at 37°C in a humidified atmosphere containing 5% CO₂ for 24 h. Cells were then treated with serial concentrations of the test samples in combination with 1 µg/mL LPS and incubated for an additional 24 h. The model group was treated with LPS alone, the blank control group received culture medium only, and the positive control group was treated with indomethacin. The final concentration of DMSO in the culture system was maintained at ≤ 0.1%. After treatment, NO production in the culture supernatant was measured using a commercial NO assay kit according to the manufacturers’ instructions.

For the cytotoxicity assays, RAW 264.7, GC‐2spd(ts) and Hep‐G2 cells were seeded into 96‐well plates at a density of 1 × 10⁴ cells/well. RAW 264.7 and GC‐2spd(ts) cells were cultured in DMEM supplemented with 10% FBS, while Hep‐G2 cells were cultured in RPMI‐1640 supplemented with 10% FBS. After incubation at 37°C in a humidified atmosphere containing 5% CO₂, cells were treated with serial concentrations of the test samples for 24 h. The final concentration of DMSO in the culture system was maintained at ≤ 0.1%. Cell viability was determined using the CCK‐8 assay according to the manufacturer's instructions. The IC₅₀ values were calculated by nonlinear regression using GraphPad Prism.

3. Results and Discussion

3.1. Ion Fingerprinting and Chemometric Analysis Reveals Extensive Overlap of Chemical Features but Distinct Overall Abundance Patterns Between RX and RB

Representative total ion chromatograms (TICs) of the RX and RB samples are shown in Figure 1. A total of 5393 and 9936 feature ions were detected in RX and RB, respectively, indicating that substantially more ion features were detected in RB. Based on the retention time, m/z values and relative peak areas of the feature ions, ion fingerprint maps were constructed for RX and RB (Figure 2A,B). Most detected ions exhibited m/z values below 1000, with ions in the range of 600–1000 being particularly abundant.

FIGURE 1.

Representative total ion chromatograms of root xylem and root bark samples.

FIGURE 2.

(A) Ion fingerprinting map of root xylem. (B) Ion fingerprinting map of root bark. (C) Principal component analysis score plot of root xylem and root bark samples. (D) Principal component analysis loading plot of root xylem and root bark samples. (E) Hierarchical clustering analysis of root xylem and root bark samples.

Given the substantial difference in the number of detected feature ions between the two tissues, a conventional bidirectional similarity index, such as the Jaccard or Tanimoto coefficients, would not directly indicate how much of the chemical feature space of the medicinally used RX was represented in RB [16, 17]. Therefore, an ion coverage metric was used to evaluate the extent to which the ion features detected in RX were also observed in RB. Specifically, RX was defined as the reference set and the coverage was calculated as the proportion of RX feature ions that could be matched in RB. This reference‐dependent metric was intended to characterise the retention of the RX‐associated chemical feature space in RB rather than to represent a symmetrical similarity coefficient. Based on this approach, 4373 of the 5393 feature ions detected in RX were also observed in RB, corresponding to an ion coverage of 81.09%. This high coverage indicates that the two tissues shared a substantial proportion of their detected chemical profiles, despite the markedly higher number of feature ions detected in RB.

The ion fingerprint data were further subjected to PCA to examine differences in the overall abundance patterns of the detected chemical features. PCA was performed directly using the feature‐ion intensity data without centring or scaling, so that the resulting principal components retained information related to the original ion abundance. As shown in Figure 2C, RX and RB samples were clearly separated along PC1, while the RX samples showed relatively tight clustering. The loading plot (Figure 2D) further identified the variables that contributed substantially to the separation between the two tissues. Combined with the compound identification results (Table S3), the major contributing variables included wilforgine (1), peritassine A (2), celastrol (3), wilforine (4), demethylzeylasteral (5), euonine (6), wilfordine (7) and wilfortrine (8). Hierarchical clustering analysis further supported the multivariate pattern observed by PCA, with RX and RB samples being clearly separated into two clusters (Figure 2E).

Taken together, the ion fingerprinting and chemometric analysis revealed two complementary aspects of the chemical relationship between RX and RB: the two tissues shared a high proportion of detected ion features, whereas their overall abundance patterns were distinctly different. These findings suggest that the differential utilisation of RX and RB may be associated not only with differences in the presence of chemical features but also, importantly, with differences in the abundance of shared constituents.

The ion coverage metric has several limitations. Because it is reference‐dependent and does not account for differences in ion abundance, the 81.09% coverage should not be interpreted as an absolute or symmetrical similarity measure. It specifically represents the proportion of ion features detected in RX that were also detected in RB. Future evaluations could incorporate bidirectional and abundance‐weighted measures to provide a more comprehensive description of the chemical relationship between the two tissues.

3.2. MSI Reveals Tissue‐Specific Distribution Patterns of Differential Metabolites in T. wilfordii Roots

T. wilfordii samples from four major production regions—ZJ, FJ, HB and HN—were selected for analysis. During method optimisation, different matrices (DHB, CHCA and THAP) and application modes (spraying vs. sublimation) were compared. DHB with spraying yielded the best response and was therefore chosen for subsequent experiments. The TICs of these regional samples were shown in Figure S1. As illustrated, the average mass spectral profiles of samples from different origins were generally similar in overall contour, though variations in ion intensity were observed.

In addition to the eight compounds that showed major contributions to the PCA separation (Figure 2D), several representative constituents of T. wilfordii were selected for MSI to further characterise their tissue localisation. These included triptolide, triptolidenol, tripdiolide, triptonide, tripchlorolide, triptriolide, triptophenolide, triptoquinone B, triptoquinone A, pristimerin and wilforlide A. The selection was based on their reported pharmacological and toxicological activities, as well as their relevance to the quality evaluation of T. wilfordii herbs and preparations. For example, tripdiolide, triptonide and tripchlorolide are representative bioactive and toxic diterpenoids of T. wilfordii, while pristimerin represents an important bioactive triterpenoid [18]. Triptolide and wilforlide A are also used as quality‐control indicators for T. wilfordii medicinal materials and preparations [19]. Their optimal adduct ions and corresponding scanning ranges for each target are summarised in Table 1. Peritassine A/euonine (b) and triptolidenol/tripdiolide (i) were analysed as two isomeric pairs because the compounds within each pair shared the same preferred adduct ion. Consequently, each isomeric pair was represented by a single ion image. Triptoquinone B was not subjected to MSI because no suitable adduct ion could be obtained under the established conditions.

TABLE 1.

Target compounds selected for mass spectrometry imaging analysis and their corresponding adduct ions and scan ranges.

No.	Compounds	Chemical formula	Optimal adduct ions	Scan range (m/z)
a	Wilforgine	C41H47NO19	[M+K]+	896.2381 ± 0.0090
b	Peritassine A/Euonine	C38H47NO18	[M+K]+	844.2432 ± 0.0084
c	Wilforine	C43H49NO18	[M+K]+	906.2588 ± 0.0091
d	Wilfordine	C43H49NO19	[M+K]+	922.2537 ±0.0092
e	Wilfortrine	C41H47NO20	[M+K]+	912.2330 ± 0.0091
f	Celastrol	C29H38O4	[M+H]+	451.2845 ± 0.0045
g	Demethylzeylasteral	C29H36O6	[M+H]+	481.2586 ± 0.0048
h	Triptolide	C20H24O6	[M+K]+	399.1211 ± 0.0040
i	Triptolidenol/Tripdiolide	C20H24O7	[M‐H2O+K]+	397.1060 ± 0.0040
j	Triptonide	C20H22O6	[M‐H2O+Na]+	363.1209 ± 0.0036
k	Tripchlorolide	C20H25O6Cl	[M+K]+	435.0978 ± 0.0044
l	Triptriolide	C20H26O7	[M+K]+	417.1317 ± 0.0042
m	Triptophenolide	C20H24O3	[M+Na]+	335.1619 ± 0.0034
n	Triptoquinone A	C20H24O4	[M‐H2O+H]+	311.1649 ± 0.0031
o	Pristimerin	C30H40O4	[M+K]+	503.2565 ± 0.0050
p	Wilforlide A	C30H46O3	[M‐H2O+K]+	475.2986 ± 0.0048
q	Triptoquinone B	C20H26O4	/	/

The microscopic images (MI), total ion (TI) images and spatial distribution maps of the target compounds are presented in Figure 3. The TI images showed generally higher overall ion intensity in the RB than in the RX region. Most target compounds also showed stronger signals in RB, although their spatial distributions differed considerably among tissues. Based on their predominant localisation, the target compounds could be broadly classified according to their predominant tissue localisation. Compounds a–e and triptophenolide (m) were predominantly localised in the phloem, whereas celastrol (f), demethylzeylasteral (g), pristimerin (o) and wilforlide A (p) were mainly enriched in the periderm. In contrast, triptoquinone A (n) was mainly detected in the xylem. Triptolide (h), triptolidenol/tripdiolide (i), triptonide (j), tripchlorolide (k) and triptriolide (l) showed broader distributions across the xylem, phloem and periderm.

FIGURE 3.

The microscopic image and spatial distribution maps of the total ion and target compounds (a–p) of T. wilfordii from four major production regions (N = 1).

When considered according to structural class, clear differences in tissue distribution were observed among the major metabolite groups. The sesquiterpene pyridine alkaloids (a–e) showed preferential localisation in the phloem, whereas the triterpenoids (f, g, o and p) were predominantly enriched in the periderm. In contrast, the diterpenoids (h–n) generally showed broader distributions across the root tissues, although individual compounds exhibited distinct localisation patterns. These observations indicate that tissue specificity was not determined solely by metabolite class and could also differ among structurally related constituents. Differences in relative signal intensities among geographical origins were also observed for some diterpenoids. For example, triptolidenol/tripdiolide (i) and tripchlorolide (k) showed stronger signals in the xylem of the Fujian sample than in the corresponding tissues of samples from other regions.

Because only one biological sample was analysed for each geographical origin, the observed differences among regions should be regarded as descriptive rather than statistically validated. Nevertheless, the major tissue distribution patterns observed in the present study were consistent with previous MSI findings in the congeneric species T. regelii. In that study, alkaloids were mainly localised in the phloem, whereas triterpenoids were preferentially distributed in the outer tissues, showing tissue‐specific distribution patterns like those observed in the present T. wilfordii samples [20]. Beyond these shared distribution patterns, the present study enabled the spatial visualisation of multiple diterpenoids across the root tissues, particularly triptolide (h), a representative bioactive diterpenoid of the genus Tripterygium. The successful visualisation of these diterpenoids provides additional spatial information on their tissue distribution and extends the MSI characterisation of diterpenoids in Tripterygium roots.

Despite the use of authentic standards and optimisation of the preferred adduct ions, the assignment of spatial distributions based on precursor ion m/z cannot completely exclude interference from structurally related or isomeric metabolites sharing the same precursor ion. This limitation is particularly relevant to the combined imaging of peritassine A/euonine (b) and triptolidenol/tripdiolide (i). Integrating ion mobility separation with MSI could provide an additional dimension of molecular separation, thereby improving the discrimination and spatial characterisation of structurally similar metabolites in future studies [21].

3.3. Targeted Quantitative Analysis Confirmed Differential Abundance of Metabolites Between RX and RB

To further characterise the differences in metabolite abundance between RX and RB, the selected constituents were quantitatively determined using UHPLC‐QQQ‐MS/MS. Because the target compounds differed substantially in concentration, polarity and ionisation behaviour, different sample dilution factors and chromatographic conditions were used to ensure reliable quantification. The overlaid MRM chromatograms are shown in Figures S2–S4 and the quantitative results are summarised in Table S4.

As shown in Figure 4A, most of the quantified compounds accumulated at significantly higher levels in RB than in RX. The exceptions were triptoquinone B (q) and triptoquinone A (n), both of which showed higher concentrations in RX. Compounds a–g were present at mg/g levels in RB, whereas most of the other constituents (h–p) occurred at µg/g levels. Tripchlorolide (k) was below the limit of quantification in all RX samples. Overall, the quantitative results demonstrated pronounced differences in metabolite abundance between the two tissues, with RB showing substantially higher levels of most of the investigated constituents.

FIGURE 4.

(A) The average content of each target compound in root xylem and root bark samples (N = 20). (B) The fold changes in average compound concentrations between root xylem and root bark samples. Positive values indicate that the compound concentration is higher in root bark than in root xylem, whereas negative values indicate that the compound concentration is higher in root xylem than in root bark. Compound k was excluded because its concentration in all root xylem samples was below the limit of quantitation, preventing calculation of the fold change (N = 3).

The magnitude of these tissue differences was further evaluated by comparing the average concentrations between RB and RX (Figure 4B). Triptonide (j), demethylzeylasteral (g) and celastrol (f) showed the largest fold differences, with concentrations in RB being 46.4‐, 44.3‐ and 36.2‐fold higher than those in RX, respectively. Triptolide (h) was also substantially enriched in RB, with a 5.3‐fold higher concentration than in RX. These quantitative differences were consistent with the stronger signals of the corresponding compounds observed in the MSI results, providing complementary quantitative evidence for their preferential distribution in RB.

Despite the clear differences between the two tissues, considerable variation was also observed among samples within the same tissue. To further characterise this variation, the range, median and interquartile range of each quantified compound were calculated (Table S5). Several constituents showed substantial dispersion and occasional extreme values among individual samples. This variation may be related, at least in part, to the fact that all samples were collected from wild plants growing under different environmental conditions, with differences in geographical location and altitude among collection sites. Such environmental heterogeneity can contribute to variation in the production and accumulation of secondary metabolites in medicinal plants [22]. Nevertheless, the overall differences between RX and RB remained evident across the quantified constituents.

3.4. Comparison of Cytotoxicity and Anti‐Inflammatory Effects Between RX and RB

The biological properties of RX and RB were further evaluated by assessing cytotoxicity and anti‐inflammatory activity. Cytotoxicity was determined in RAW 264.7, GC‐2spd(ts) and Hep‐G2 cells (Figures S5–S7). Some samples did not reach 50% inhibition of cell viability within the tested concentration range; therefore, their IC50 values were estimated and indicated with a tilde (∼). Across all three cell models, RX consistently exhibited lower cytotoxicity than RB (Table 2 and Figure 5A), indicating that RX was less cytotoxic than RB under the tested conditions.

TABLE 2.

IC₅₀ and EC₅₀ values in different cell models of root xylem and root bark samples (N = 3).

		IC50 (µg/mL)			EC50 (µg/mL)
	No.	RAW 264.7 cell viability	GC‐2spd(ts) cell viability	Hep‐G2 cell viability	NO inhibition in RAW 264.7
RX	RXA	∼ 894.1	134.6 ± 2.599	∼ 272.1	250.1 ± 19.95
	RXB	∼ 584.8	89.13 ± 2.950	378.2 ± 42.77	71.64 ± 6.587
	RXC	∼ 765.1	106.2 ± 3.466	170.3 ± 13.30	226.1 ± 23.61
RB	RBA	130.9 ± 6.600	3.004 ± 0.103	40.02 ± 3.619	35.5 ± 2.695
	RBB	98.85 ± 11.66	1.813 ± 0.185	21.5 ± 1.750	22.53 ± 0.819
	RBC	101.2 ± 1.403	2.712 ± 0.241	23.22 ± 2.948	16.31 ± 2.042

FIGURE 5.

(A) IC₅₀ ranges of root xylem and root bark for cytotoxicity in RAW264.7, GC‐2spd(ts) and Hep‐G2 cells (N = 3). (B) EC₅₀ values of root xylem and root bark for NO inhibition, with indomethacin used as the active control (N = 3). (C) Correlation between content of bioactive compounds and IC₅₀ of cell lines. Compound k was excluded from the correlation analysis because its concentrations in all root xylem samples were below the limit of quantitation.

The anti‐inflammatory activities of RX and RB were assessed based on inhibition of LPS‐induced NO production in RAW 264.7 cells (Figure S8). RB showed markedly stronger NO‐inhibitory activity, with EC₅₀ values ranging from 16.31 to 35.50 µg/mL, compared with 71.64 to 250.1 µg/mL for RX (Table 2). The inhibitory activity of RB was not significantly different from that of the active control (AC), whereas RX showed significantly weaker activity than AC (p < 0.05), as shown in Figure 5B.

To explore the relationship between chemical composition and biological effects, correlation analysis was performed between the contents of quantified compounds and the measured cytotoxicity and anti‐inflammatory activity (Figure 5C). Strong positive correlations (r ≥ 0.6) were observed between the levels of peritassine A (b), wilforine (c), demethylzeylasteral (g), triptolide (h), triptriolide (l) and triptophenolide (m) and the measured biological endpoints. Among these compounds, demethylzeylasteral (g), triptolide (h), triptonide (j), triptriolide (l) and triptophenolide (m) have been previously reported to exhibit direct anti‐inflammatory activity or cytotoxicity [1, 23, 24, 25, 26]. For compounds such as peritassine A (b), for which literature evidence regarding bioactivity or toxicity remains limited, the observed correlations provide candidate associations for further investigation.

3.5. Chemical and Biological Basis for the Distinct Applications of RX and RB

The chemical and biological findings obtained in this study provide a possible basis for understanding the differential utilisation of RX and RB. Rather than differing primarily by the presence or absence of completely distinct metabolite classes, the two tissues shared a substantial proportion of detectable chemical features but differed markedly in the abundance and spatial distribution of several constituents. In particular, RB showed greater accumulation of several constituents associated with both pharmacological activity and cytotoxicity, which was accompanied by stronger anti‐inflammatory activity but also higher cytotoxicity. In contrast, RX exhibited lower abundance of many of these constituents and lower cytotoxicity, although its anti‐inflammatory activity was also weaker. These findings suggest that the distinct biological properties of RX and RB may be related to differences in the levels and tissue localisation of shared and tissue‐enriched constituents, rather than simply to differences in chemical composition.

This interpretation is consistent with the historical and contemporary utilisation of T. wilfordii. RX has traditionally been more commonly used for internal administration and remains the principal raw material for standardised preparations such as Tripterygium glycosides tablets, whereas RB has been used primarily for external applications [3, 4, 5]. The lower cytotoxicity observed for RX is consistent with its continued use in internally administered preparations, while the stronger cytotoxicity observed for RB may partly explain its more restricted traditional use. Together, these findings suggest that the differential utilisation of RX and RB may reflect their distinct efficacy‐toxicity characteristics.

From the perspective of quality evaluation and resource utilisation, these findings also have practical implications. Because RX and RB share many detectable chemical features while differing substantially in the abundance and spatial distribution of key constituents, evaluation based solely on the presence or absence of individual marker compounds may not fully reflect the chemical and biological characteristics of the starting material. The marked abundance differences identified in the present study therefore warrant further investigation of the pharmacological and toxicological activities and underlying mechanisms of these constituents. Clarifying how these abundance differences contribute to the efficacy and toxicity of RX and RB will be an important step toward establishing tissue‐relevant quality markers and defining appropriate quality‐control strategies for T. wilfordii [27, 28]. Such evidence will also be important for determining the medicinal value and appropriate utilisation of different root tissues.

However, the chemical–biological relationships identified in this study should be interpreted within the scope of the experimental evidence. The observed correlations between individual constituents and biological endpoints indicate associations rather than direct causality, particularly because the biological effects of complex plant materials may result from the combined actions of multiple constituents. In addition, the present cell‐based assays do not establish the molecular mechanisms underlying the differential cytotoxicity or anti‐inflammatory activity of RX and RB. Further studies using purified constituents, mechanistic cellular models and appropriate in vivo or pharmacological approaches will therefore be required to determine the individual and combined contributions of the identified constituents.

4. Conclusion

This study integrated ion fingerprinting, chemometric analysis, MSI, targeted quantitative analysis and biological evaluation to investigate the chemical and biological differences between the RX and RB of T. wilfordii. Ion fingerprinting and chemometric analysis revealed substantial overlap in their detected chemical features but distinct overall abundance patterns, while MSI and targeted quantitative analysis further resolved differences in metabolite localisation and abundance. These chemical differences were accompanied by distinct biological properties, with RB showing greater accumulation of several bioactive constituents, including triptolide, triptonide and celastrol, together with stronger anti‐inflammatory activity but higher cytotoxicity than RX.

The findings indicate that the distinct characteristics of RX and RB are not primarily attributable to completely different chemical compositions, but rather to differences in the abundance and spatial localisation of shared and tissue‐enriched constituents. The greater accumulation of several bioactive constituents in RB was associated with both enhanced anti‐inflammatory activity and increased cytotoxicity, whereas the lower abundance of these constituents in RX was accompanied by lower cytotoxicity but also weaker anti‐inflammatory activity. This efficacy–toxicity difference provides a possible chemical and biological basis for the distinct traditional and contemporary utilisation of the two root tissues.

These findings also highlight the importance of considering tissue composition, metabolite abundance and associated efficacy–toxicity characteristics when evaluating the quality and utilisation of T. wilfordii root materials. Further investigation of the pharmacological and toxicological activities and underlying mechanisms of the markedly differential constituents will be important for establishing tissue‐relevant quality markers and supporting the rational utilisation of different root tissues.

Author Contributions

Zhong Mou Zhang: writing – original draft, investigation, formal analysis, data curation and methodology. Kai Ling Xu: investigation and writing – review & editing. Qing Zhang: formal analysis and data curation. Yong Qiang Lin: writing – review & editing, project administration, supervision and conceptualisation. Feng Wei: writing – review & editing, project administration, supervision and conceptualisation. Ya Dan Wang: methodology, formal analysis, writing – review & editing and funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding

This work was supported by the Training Fund for academic leaders of NIDC (2025×2).

Supporting information

Supporting File: ansa70112‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.